Feasibility study of networks. Progress report, July 15, 1975--July 14, 1976

The Laboratory for Nuclear Science has continued its study into the feasibility of networks, changing the focus of the investigation from the possible ways of collaborating with the MIT Information Processing Center in implementing a connection to the ARPA net to a wider consideration of the implications of computer resource sharing via networks. This report describes experiments performed on the ARPA net in collaboration with the Lawrence Berkeley Laboratory and experiments in more conventional forms of remote job entry to that site and others in ERDA. It discusses the use of the Multics mail as set up for ERDA, comparing it to other mail systems. Plans for future experiments in the ARPA net and for investigation of networks other than the ARPA net are proposed. The conclusion was reached that resource sharing via networks is both feasible and desirable provided certain interface problems are solved. Recommendations are made for more investigation into the human factors involved in network use, methods of providing consulting help to users, and determination of the most cost effective use of remote resource sharing. For the Laboratory for Nuclear Science, it is concluded that access to other ERDA computers via networks will meet some of our computing needs now and more in the future. However, remote resource sharing can be used only to supplement the on site computer, not replace it. At the present time, the best method of accessing the ARPA net for our users is via the MIT Multics System. However, as a permanent solution, a mini-host located in our computer center may provide more capability. In the next year, we plan to work out methods of attacking the problems of interfacing to the ARPA net and to use it in a limited manner for actual production work